Monday, 5 August 2013

Plagiarism, Lies and Bullshit - Part 2, the "Straw Man" exposed

In Part 1 I summarised results for 85 australian stations over the period 1990-2010, which the authors of "Present-to-future sea level changes: The Australian case" appear to think comprises 20 years.

For anyone who thinks I'm just moaning about a few "little mistakes" that Mörner & Parker have made (it's actually written by Mörner alone, as I'll explain later), I'll state what I've found in this "paper" - all that's in my post title. Mörner lies about his results; not selective reporting or creative statistics, simple straightforward lies. Lies that are very easy to disprove; anyone with access to the internet and with no knowledge of statistics is able so see they're lies. Disinformation, and its partner in crime, unadulterated bullshit, is present throughout, like some constant and annoying in-store muzak. The whole point of this rotten pamphlet is to attack a "straw man", aided and abetted by blatant plagiarism.

Anyone perpetrating a fraud, the representation of fiction as truth, cannot maintain a consistent story without inadvertently revealing apparently small but significant contradictions and disconnects; this "paper" contains many such. I will detail some of the most obvious, which seem to have escaped the notice of surprisingly un-sceptical and credulous "sceptics", both bloggers and comment posters who would delight in tearing apart any paper or blog post from those on "the other side" which contained so little in the way of actual results. This paper contains not even one, so I'll provide them in a full analysis of the entire 86 (actually 85) stations which Mörner and Parker claim to have analysed over the 21 years 1990-2010, which they appear to believe spans 20 years.

There's clear confusion throughout as to what constitutes periods of 10 and 20 years - a kind of "date dyslexia". The date ranges which are claimed to have been analysed are given as 1990-2010, 1990-2000, and 2000-2010. These are periods of 21, 11, and 11 years respectively, though the text mentions "20 years", "the last 10 or 20 years", the introduction "the last two decades", and the caption for the only chart included says "for the last 10 and 20 years". The short ranges are not consecutive; they overlap at 2000. With such confusion evident, one might ask exactly which periods were supposed to have been analysed - were they analysed?

Exaggeration - but by whom?

First, I'll rebut the the accusation of "exaggeration" in the "official Australian governmental value" of 5.4 mm/year. You won't find that figure anywhere but in this paper, or in blog posts discussing it. Google it if you will. It's a simple average of the individual ABSLMP station trends to June 2011 which were tabled in an ABSLMP report (June 2011, published Sept. 2011, no overall average included), and as such cannot be "exaggerated". It may not be exactly relevant to the long-term sea-level change around Australia, but that's another matter entirely, one which I'll cover later.

Two stations in South Australia are uniquely close to one another; Port Adelaide and Port Stanvac (an ABSLMP station) are just 20 km apart. The latter was dismantled after Dec. 2010, when the adjacent oil refinery was closed and decommissioned. Their records, over the period of overlap, differ by an average of 190mm, due to different "tide-gauge zero" benchmarks in use. All tide-gauges have local benchmarks; the gauge readings are not absolute but relative to the local benchmark. Across Gulf St. Vincent, about 60 km to the west is Port Giles, though you won't find it on any map; it's a jetty and a few buildings. Its gauge record differs from Port Stanvac by some 50 mm. Using the Port Stanvac ABSLMP record (from June 1992 to Nov, 2010) as a baseline, I've plotted the adjusted record from Port Adelaide (Jan 1990 to Dec 2010) on the same chart. I repeated this for the shorter Port Giles record, Sept. 1994 to Dec. 2010.

The degree of overlap and correlation is remarkable, I'd say, much better than I'd expected. Two things are evident - that the ABSLMP station very closely reflects sea-level at the two other Gulf stations, and that the rate for Adelaide greatly exceeds the long-term rate of about 2.3 mm/year. I say "about", because an NTC report gives it as 2.2 mm/year to 2009 (more , much more, later), and the latest data to 2012 gives 2.35 mm/year; here's the chart.

Three things are evident; the moving average shows sea-level at Port Adelaide rose only a little between 1965 and 1990, there was a dip to about 1996, and after that rose more sharply than the overall trend. That's a general profile for most Australian stations, though to a greater degree for many, and to a far lesser degree for some. The rate for the 1990-2010 period for Port Adelaide is not much greater than the average for all stations, as you'll see. Here's the chart fo Fremantle, WA to 2012. The profile 1940 to 2012 is almost identical; it's the Australian story over that period, and both charts blow the "no acceleration" meme apart. The acceleration isn't due to melting glaciers or AGW in general, it's simply returning Oz sea-levels to above "normal" after decades of ENSO-induced variations, including the early-mid 1990s "dip", the result of two El Ninos; 1991-2 and 1994-5.

Much greater distances separate Wyndham, Western Australia, from Broome (ABSLMP), and from Darwin, Northern Territory (also ABSLMP); about 500 and 800 km respectively, I'd say. I've repeated the exercise for these three stations - the necessary adjustments are shown on the charts.

It's said a picture's worth a thousand words; a well-chosen moving average is worth any number of "2nd. degree polynomials", because it focusses on what's been (and is) going on.

It's now clear where the "exaggeration" originates - Mörner & Parker, henceforth referred to as "M&P". It's also clear that their claim of "No fitting produces a sea level rise in excess of 1.5 mm/year" (which would have to include these three stations) is totally demolished (although they do that rather well themselves - more later!). Of course, as the 3rd. Viscount Monckton of Brenchley might say (and has done, many times) "Don't believe a word I say" - check for yourselves; I provide NTC and PSMSL links for all stations I've analysed. His instruction also applies to M&P of course - why do so many otherwise (apparently) intelligent people believe everything they say without checking anything? Do they (particularly Mörner), have some sort of "free pass", which makes what they say and write immune from scrutiny? Isn't that kind of chauvinism both unscientific and un-sceptical, in short, partisan, exactly what sceptics criticise "the other side" for? Surprisingly, the noble lord is a great fan of Mörner's; perhaps he should pay more attention to what he instructs others to do, and be a little more sceptical himself. Rant put on hold - now for some more real data.

The actual and factual analysis

A number of stations have so little data for the relevant period they don't bear any analysis. One station has its record duplicated on the NTC page for Queensland. Strange therefore that these inconvenient facts weren't mentioned. Neither mentioned is that the remaining 85 stations include 6 which are thousands of km away from Australia; one is halfway between New Zealand and Antarctica, 4 are on the Antarctic coast, and one (Cocos Islands) is best described as being half-way between Darwin and Sri Lanka, in the Indian Ocean, and closer to the Maldives than Darwin. Relevant? Of course not.

My analysis in Part 1 is open, comprehensive, detailed and checkable, but does it prove anything about future trends in sea-levels around Australia? No, it does not; the period 1990-2010 starts at a low point in the record. M&P missed that, but it should, and would, have been obvious to anyone who'd charted the data and added meaningful moving averages. Sea-levels generally declined between the 1960s and 1993-4; any trends calculated from the early 1990s to present reflect the ENSO-driven rebound from that low period. Over the next 5-10 years, both the decline and sharper recent rise will "average out" as the record lengthens. There has indeed been an acceleration in rates in the last two decades; anyone who claims otherwise is being disingenuous at best. Anyone who claims that this unrepresentative period "proves" that the acceleration is caused by anything other than ENSO-induced change superimposed over a relatively constant underlying trend is being more than disingenuous.

In a recent post, I showed the correlation between the Southern Oscillation Index (SOI) and sea-level at two Australian, and several Pacific island stations; I could have cited a number of other examples, both from Australia and Pacific islands. I even attempted to remove the ENSO effect, and succeeded IMHO. At least all the "lumps and bumps" (and dips) were evened out. ENSO is the "big player" in the Pacific - underlying rates of rise are swamped by the ENSO signal. Anyone who ignores ENSO (as measured by the SOI) are fated to draw entirely wrong conclusions about what has been happening, or is happening, or will happen wrt sea-levels in the Pacific in general and around Australia in particular, in the context of this and my last post.

If I seem to be "going on a bit" about ENSO, and "over-egging the pudding", I don't apologise - it's important, very important in this context, and not just a few dips and bumps in the records.

Major disconnects in M&P's "results*

Here's the list of claimed analyses and results for Australian stations:

70 "non-ABSLMP" stations 1990-2010, linear trends: average 0.1 mm year
2nd. degree polynomials for same: none
70 "non-ABSLMP" stations 1990-2000 linear trends: none
2nd. degree polynomials for same: none
70 "non-ABSLMP" stations 2000-2010: none
2nd. degree polynomials for same: none
16 ABSLMP stations 1990-2010 linear trends: none
2nd. degree polynomials for same: none
All 86 stations 1990-2010, linear trends: average 1.5 mm year
and "No fitting produces a sea level rise in excess of 1.5 mm/year".

Now if 86 stations produce an average of 1.5 mm/ year, and 70 stations an average of 0.1 mm/year, it should be clear the 16 ABSLMP stations must have contributed rather a lot to the average; they comprise only 19% of the total. To boost the average from 0.1 to 1.5 they must have averaged very much more than 1.5. Remember "No fitting produces a sea level rise in excess of 1.5 mm/year"? That average is central to M&P's claims, yet we're not informed what it was? However, it's easy to work out the missing value for the 16 ABSLMP stations:

86 x 1.5 gives a cumulative total of 129; the 70 stations give a cumulative total of 7, so the 16 stations contributed 122, giving an average of 7.625 mm/year. The apparently "exaggerated" NTC average quoted was just 5.4 mm/year - where does the 7.625 figure come from? I thought "No fitting produces a sea level rise in excess of 1.5 mm/year"? Both station averages are fiction - the ABSLMP average to 2010 (from the NTC 2010 December report) is 4.9 m/year, and if that were used, the 70 stations would have an average of (129 - 78.4)/70 or 0.72 mm/year. In the Discussion section, Morner says
From this comparison it seems obvious to us that the Australian governmental value of 5.4 mm/year must be significantly exaggerated. The Australian data analysed by us provide a range from 0.1 to 1.5 mm/year. The same over-estimation seems to apply for individual sites when comparing our values from Darwin (their 8.6 versus our 2.2 mm/year) and from Stony Point (their 2.6 versus our -2.1 mm/year).
So the "Australian governmental" value is "exaggerated" at 5.4 mm/year, but his average figure is 7.6? I remind you that my properly derived average is 4.35 mm/year - that doesn't mean that the NTC figure is incorrect, as their trends were calculated to June 2011, rather than to Dec. 2010, start in various years from 1990 to 1993, and have a residual of a half-year of the annual cycle. Mine were calculated over the full 1990-2010 period, and so include earlier data for the 14 gauges which replaced existing installations.

And do we get some actual results at last? No we don't - the "comparisons" are sleight-of-hand - smoke and mirrors. "Their" figures for Darwin and Stony Point are for the entire length of record, not 1990-2010, and it's worth while pointing out that longer-term records for the 14 of the 16 stations contain ABSLMP data - the gauges were replaced. The other two are all ABSLMP data, so there can be no difference there either. The NTC simply extended the 14 stations with data from the new gauges.

Links to sea-level data for all charts can be found on the NTC data page.

I'll update this post with charts for Darwin and Stony Point, and direct links to data later - I wanted to get some actual charts displayed, to confirm that the "exaggeration" claim was false, and that the "No fitting produces a sea level rise in excess of 1.5 mm/year" was also false. Watch this space

No comments:

Post a Comment